Skip to main content Skip to footer

Our work of tomorrow is going to look vastly different from the work we do today. AI is augmenting workers in certain roles, and soon this trend will expand to the vast majority of the workforce, allowing humans to do the work that matters and, in the process, vastly improving workforce productivity and effectiveness.

But rather than thinking of AI as a superpower in and of itself, it’s essential to view intelligent systems as augmenting human work. The inherent power of AI lies in its ability to deliver the “science” of work while humans complete the “art” of work. This symbiotic relationship is what will define work moving forward.

A new way to approach work

To understand which tasks should be allocated to intelligent machines vs. humans, organizations need to see work not as a broadly defined set of work functions (e.g., research, quality control, product development, etc.) but as a collection of individual activities (i.e., physically connecting wires in a manufacturing process, reporting irregularities to a legal team, logging parts received in a shipment). By analyzing the individual components of job roles, organizations can better assess which are best suited for human workers vs. intelligent systems.

One way to do this is to categorize the functions or tasks within a job role into two buckets, as we did in our previous study “Space Matters: Shaping the Workplace to Get the Right Work Done”:

  • Red tasks. Rote, repetitive work with little need for creativity but potentially high amounts of calculation and attention to detail. Because these tasks involve complex data analysis, repetition, pattern recognition and a low level of human interaction, they’re better performed by AI. Think of document scanning, individual parts logging, customer data entry or highly structured risk analysis.
  • Blue tasks. Collaborative, social work with high business impact. Because these tasks demand judgment, empathy and creativity, they’re better performed by people than machines. They rely on people gathering together to iterate, experiment, discuss, create and innovate. Think of engagement with an irate customer, corporate networking or exception handling.

Organizations can tune the interplay of humans and machines by applying AI to “red” tasks, thus facilitating workers to better focus on their “blue” tasks. We call this the “task master model.”

Physical task master model

We’ve designed two task master frameworks, one for cognitive and one for physical tasks.

Within the physical task master model, we analyze tasks according to two criteria: the relative social interaction that each entails and the dexterity and structure that each requires. 

The best way to understand this is through an example, such as Boeing’s implementation of AR headsets for its wiring loom workers. Traditionally, workers had to refer to massive manuals to reference-check thousands of connections in a wiring loom. The constant toggling between laptop and wiring work caused immense eyestrain and poor productivity.

Boeing introduced AR headsets equipped with natural language processing capabilities. Workers can now ask the headset to describe next possible connections, with the response delivered via an AR interface or hands-on video. This has reduced the production time of wiring looms by 25%.

Seen through the task master model, a low-dexterity, highly structured task that involves no human interaction includes parts logging. This is, therefore, a prime task to be automated. However, the physical connection of wires to the harness, while not requiring any social interaction, does require a high degree of dexterity and, therefore, remains the role of the human worker for the time being. Discussion with managers, colleagues and suppliers, however, is and will remain a human task.


The red and blue designations within each quadrant provide guidance on the nature of the task, with blue representing human tasks and red representing tasks that can be automated. The bottom right quadrant is both red and blue, as tasks within this quadrant are not yet automatable, but over time, this will slowly change as intelligent machines become more creative and able to understand a greater number of contextualized variables, therefore becoming more strategic.

Figure 1

The cognitive task master model

Within the cognitive task master model, we analyze tasks according to three criteria: the social interaction each task entails, the creativity and strategy required and whether the task is transactional or oriented toward optimization.

Let’s look at how this works at a global professional services organization that wanted to improve its risk management due diligence process. International due diligence involves exhaustive research, with more than 40,000 global sources tracking not only media but also corporate records, financial transactions and legal cases. Results based on analysts’ text strings needed to be painstakingly reviewed for each entity before a report could be finalized.

By using application programming interfaces (APIs) to link the company’s due diligence software to a machine-learning interface, the system was able to learn many of the researchers’ tasks, thereby drastically reducing research time, with 14% of reports complete in one hour. Generation of due diligence reports also increased by up to 30% per year, with increased accuracy.

In this case, due diligence is a multi-faceted role that could not be entirely automated. However, due to the high degree of accuracy required and the scale to which that accuracy is necessary, it’s a prime example of how AI can augment the cognitive worker. Tasks such as stakeholder engagement and irregularity appraisal are complex, strategic and social; therefore, these will remain purely human tasks. Irregularity reporting to legal, while relatively transactional in nature, is highly social due to the conversations required to discuss the intricacies of a specific case.

AI is particularly well-equipped, however, to handle tasks such as word indexing, document scanning and exception risk handling, which it can execute at a scale and accuracy that humans could not.


The red and blue designations within each quadrant provide guidance on the nature of the task, with blue representing human tasks and red representing tasks that can be automated. The bottom right quadrant is both red and blue, as tasks within this quadrant are not yet automatable, but over time, this will slowly change as intelligent machines become more creative and able to understand a greater number of contextualized variables, therefore becoming more strategic.

Figure 2

HR’s retooling imperative

The role of human resources (HR) in all of this cannot be underestimated. HR teams need to look at themselves through the lens of the task master model, identifying areas within their roles that can be augmented through the use of AI systems.

Rote tasks such as leave calculation and benefits administration will erode as more emphasis is placed on purely human tasks such as compensation analysis and strategic workforce planning. In addition, certain strategic HR tasks will be augmented, including learning/development and recruiting.


Rote tasks in the HR value chain that can be automated through AI systems are represented in red. Tasks that can be augmented through AI are in orange. Purely human tasks are in blue. Tasks that will need to be implemented within HR are in green.

Figure 3

Organizations around the world are reaching a tipping point: New digital technologies, including AI, are changing the dynamic of how workers deliver value. To facilitate human-AI augmentation, organizations need to hone the new discipline of objectively viewing job roles as collections of tasks. The extent of this augmentation cannot be taken lightly as the potential productivity improvements will prove to be massive.

To read the full report, see “The Symbiosis of Humans and Machines: Planning for Our AI-Augmented Future,” visit us at the Cognizant Center for the Future of Work, or contact us.